Imagine that you have a personal assistant always available, capable of correcting your writing instantly, searching for information in the blink of an eye, or even suggesting ideas for your next project.
Its name is ChatGPT and, since its conception by the OpenAI ecosystem, many can no longer imagine life without it. But what happens when we delegate too much to this artificial intelligence?
A recent MIT study suggests that our brain could take too long a break, and not exactly on vacation.
What did MIT do and how did they measure it?
Dr. Nataliya Kosmyna and her team recruited 54 university students, divided into three groups: a ‘brain only’ group, another with access to a search engine, and a third with the help of ChatGPT.
During several sessions, participants wrote entrance exam-style essays while wearing electroencephalography (EEG) headsets that measured neural activity at different frequencies.
The goal was not just to quantify words per minute, but to discover what was happening in their brain networks.
Amazing results
The data showed that those who trusted ChatGPT showed up to 55% less neural connectivity compared to the group without tools and 48% less compared to the group that used a search engine.
This disconnection translated into a more superficial encoding of the information: less memory of the content and a feeling of “not feeling one’s own” of what was written.
In other words, just as training with only light weights can atrophy your strength, delegating to AI seems to numb our cognitive muscle.
Limits that invite caution
Before proclaiming the end of human intelligence, it is vital to qualify. The study, published as a preprint, has not yet passed peer review. The sample is relatively small and the tests focused exclusively on essay writing.
Would these results be repeated in other tasks, such as solving mathematical problems or designing a marketing plan? The answer is yet to come.
Psychosocial and emotional effects
The interest in understanding how AI affects us is not limited to brain activity. Another line of research from the MIT Media Lab itself explored how the way we communicate with a chatbot (whether by text, with a neutral voice or with a more attractive intonation) influences our perception of company.
Far from measuring brain waves, this study pointed out that interacting with these digital voices can reduce the feeling of loneliness, but also generate a certain emotional dependence.
Working memory and executive functions
An extensive review of publications on MDPI (a portal to access scientific publications) looked at how chatbots affect attention and working memory.
The findings were mixed: in controlled scenarios, AI improved our ability to reorganize information or focus on fragmented tasks.
However, these benefits did not always translate to real-world environments, where distractions and time pressure add unexpected variables.
Chatbots in the office and in the classroom
Support at work
In a recent study published in Nature, the authors examined the use of ChatGPT in professional contexts.
They discovered that the “almost human” personality of these systems and their continuous training could accelerate the learning of new tools or procedures.
However, they warned that indiscriminate use could create a false sense of competition, where the worker places more trust in the AI than in their own judgment.
A virtual teacher with caveats
On the other hand, surveys of undergraduate students show that more than a third already use ChatGPT to structure assignments or resolve doubts.
Many value the time savings and clarity, but a similar number acknowledge fear of losing the ability to critically analyze. The approach is clear: if the chatbot does 80% of the work, how do we exercise that remaining 20% that forges deep understanding?
The paradox of “cognitive atrophy”
In an article published on PubMed Central, some researchers coined the term “AI chatbot-induced cognitive atrophy” (AICICA). They compare this phenomenon with the “digital laziness” that arises from stopping memorizing data because “everything is on Google.”
Overreliance on AI could erode our mental flexibility, making us less able to improvise or take on new challenges. And, perhaps most worrying, these alterations could be established without us realizing it.
Keys to a balanced use of chatbots
Should we give up these virtual assistants? Nothing could be further from the truth: chatbots are extremely powerful tools, as long as we learn to use them without neglecting our skills. Some practical recommendations:
- Purpose first: reserve the use of AI for mechanical or first-approach tasks, and leave room for creative exercise.
- Conscious alternation: alternate sessions without digital help, so that your brain continues “training”.
- Post-reflection: after using the chatbot, review what you learned in your own words to reinforce retention.
With these habits, we avoid falling into the trap of comfort and keep the internal dialogue that drives thought alive.
What can we conclude from these studies?
The MIT study opens the door to legitimate alarms: ceding too much ground to AI can translate into less mental effort, superficial learning, and loss of critical autonomy.
However, other works point to opportunities: in environments where information is enormous, chatbots can filter out noise and help us focus on what is essential.
The key lies in the context and the degree of dependency. For routine tasks, AI is a brilliant ally; When we address creative or high-level cognitive challenges, the true protagonist must continue to be our own brain.
Only in this way will we achieve a balance between technological efficiency and the development of our most authentically human capabilities.
This post is also available in: